BROWSE
Detail
Conference Paper
Trajectory Segmentation based on Spatio-Temporal Locality with Multidimensional Index Structures
Cited 1 time in 
Share 
Share
- Authors
- Yongjin Kwon, Junho Jin, Jinyoung Moon, Kyuchang Kang, Jongyoul Park
- Issue Date
- 2016-10
- Citation
- International Conference on Collaboration Technologies and Systems (CTS) 2016, pp.212-217
- Language
- English
- Type
- Conference Paper
- Abstract
- Despite the remarkable growth of video analysis technologies, human operators still suffer from the difficulties of careful monitoring of a lot of videos in many industrial applications. Since a number of methods for understanding videos usually consider object movements, it is also concentrated on trajectory analysis. Due to the high and variable dimensionality of trajectories, trajectory analysis! not trivial. Some studies divided each trajectory into several pieces. However, the lack of discussions on how to segment concerning trajectory analysis led to flood too naive or too complicated methods. In this paper, we propose a simple but effective method of trajectory segmentation concerning sjaito-temporal locality. Using multidimensional index structures and some temporal concerns, a great set of trajectory segments can be constructed in a short time. In addition, we extracted semantic regions, as an example of trajectory analysis, with the results of trajectory segmentation. The experiments showed that trajectory segments reflect on the spatio-temporal locality, and semantic regions were well extracted, which indicated that our segmentation had potential for trajectory analysis.